A predictive surrogate model for hemodynamics and structural prediction in abdominal aorta for different...

Area of Science:

• Biomedical Engineering
• Computational Fluid Dynamics
• Medical Imaging

Background:

• Accurate prediction of blood flow and vessel mechanics is crucial for diagnosing cardiovascular diseases.
• Traditional Fluid-Structure Interaction (FSI) models are computationally intensive.

Purpose of the Study:

• To develop and evaluate a Predictive Surrogate Model (PSM) for abdominal aorta analysis.
• To integrate Proper Orthogonal Decomposition (POD) and Long Short-Term Memory (LSTM) for efficient prediction.
• To assess the model's performance under rest and exercise physiological conditions.

Main Methods:

• Utilized a Fluid-Structure Interaction (FSI) solver as the Full-Order Model (FOM).
• Employed Proper Orthogonal Decomposition (POD) for dimensionality reduction.
• Integrated Long Short-Term Memory (LSTM) networks for surrogate modeling.
• Evaluated reconstruction errors for velocity, pressure, Wall Shear Stress (WSS), and Oscillatory Shear Index (OSI).

Main Results:

• Reconstruction error for velocity components was below 10% for both rest and exercise conditions.
• Pressure prediction error was minimal, less than 1%.
• The model demonstrated consistent error magnitudes for velocity and wall-related indices.

Conclusions:

• The Predictive Surrogate Model (PSM) shows promising accuracy for predicting fluid and solid variables in the abdominal aorta.
• The PSM offers an efficient alternative to computationally expensive FOMs for physiological simulations.
• The model's performance is robust across different physiological states, including rest and exercise.